Constraint-Directed Backtracking

We propose a new backtracking algorithm called constraint-directed backtracking (CDBT) for solving general constraint-satisfaction problems (CSPs). CDBT and the naive backtracking (BT) share a similar style of instantiating and re-instantiating variables. They diier in that CDBT searches for an assignment to variables in a variable set from the given constraint posed on that variable set and appends it to an existing partial solution, whereas BT searches for an assignment of one variable from its domain. CDBT has a more limited search space and it actually visits fewer nodes than BT. The similarity between CDBT and BT enables CDBT to be incorporated with other tree search techniques such as backjumping or forward checking and consistency techniques such as the !-consistency algorithm to improve its performance further.